TERMINAL CONTROL AREA COMPLEXITY MEASUREMENT USING SIMULATION MODEL

: Traffic density in the terminal control area will increase flight safety risks. One effort to reduce the risk is to minimize the controller’s workload when affected by air traffic complexity. This research uses a simulation model to measure air traffic complexity in terminal control areas. The aircraft performance model has been constructed from ADS-B data and represents the aircraft movement in the terminal control area of Soekarno-Hatta International Airport. The simulation model can detect and resolve conflicts to keep separations between aircraft at a specified minimum separation limit. Air traffic complexity measurement uses several indicators, i


INTRODUCTION
The Terminal Control Area (TCA) is airspace with the most complex and dense system compared to other airspace sectors.Three modes of flight operations run simultaneously: arrival, departure, and cross-flight [1].TCA is highly sensitive to changes in traffic, weather, flight procedures, runway used, and other unusual events.The assessment of the system's performance is important and is affected by the system's complexity [2].
The air traffic complexity rate can be very high due to the traffic intensity and patterns in mutual interactions between different traffic flows and individual aircraft in the TCA.The increasing complexity of the TCA will increase the complexity of controller tasks and result in increased workload [3].The management of traffic flow and airspace can be carried out correctly to avoid excessive controller workload if the measurement and prediction of air traffic complexity can be carried out accurately [4].It is necessary to consider the interactions between individual aircraft and their flight characteristics to determine complexity more precisely.The interactions involve possible conflicts and the tendency for aircraft movement to converge at one point.[5].There are at https://doi.org/10.31436/iiumej.v24i1.2223least nine references from Mogford et al., Diaconu et al. and Dervic and Rank (2 that mention various indicators of complexity [4,6,7].Table 1 shows the summary of some complexity indicators.
Air traffic complexity can be measured by experts, who have experience controlling air traffic under various conditions, or by the complexity indicator obtained from air traffic data and the number of interactions between aircraft in a particular sector [16].Another method to determine air traffic complexity is using a dynamic weighted network model.The nodes represent aircraft, waypoints, and airways in the network model.The total weights of all network edges represent air traffic complexity [17].Andrasi et al. used an artificial neural networks model to estimate air traffic complexity.The best configuration of artificial neural networks was determined by a genetic algorithm [18].This research developed an ATM model simulation in the TCA using MATLAB to reflect aircraft movements and the air traffic control process.The simulation model is then used to analyze air traffic complexity in the TCA, specifically at Soekarno-Hatta International Airport, Jakarta.The simulation model has several advantages in describing the air traffic system and its complexity, i.e. the movement of the aircraft can be visualized to analyze and validate it [19].The simulation parameters also can be modified easily to obtain various scenarios.

System Description
The ATM system model represents arrival and departure operations for Runway 25R and 25L on Soekarno-Hatta International Airport -Jakarta (JAKARTA).The arriving aircraft will enter the TCA from the en-route airspace through the transition points.The aircraft then fly towards the runway following a specific trajectory profile defined by some waypoints.The arrival trajectory profile refers to Standard Terminal Arrival Routes (STAR).The departing aircraft enter the TCA from the aerodrome control area and fly to the airway following the Standard Instrument Departure trajectory profile (SID).The information about STAR and SID can be accessed in the Aeronautical Information Regulation and Control (AIRAC) as a supplement to the Aeronautical Information Publication (AIP) published by the Directorate General of Civil Aviation (DGCA) of Indonesia [20].
With Flight Management Computer (FMC) support, aircraft can automatically fly along with these profiles.Under certain circumstances, the aircraft must follow the instruction from the air traffic controller containing the aircraft's direction (heading), altitude, and speed changes, often referred to as vectoring.Arrival operations have more significant conflict potential than departures because aircraft have trajectories that converge, especially when entering a merging point.Aircraft speed on arrival will also experience a reduction so that the aircraft in front tends to be overtaken by the aircraft behind it.The departure model has a smaller potential for conflict than the arrival because the trajectory tends to be diverging.Potential conflicts with arriving aircraft are also minimal due to differences in altitude and flight path.
There are six TCA sectors related to Runway 25R and 25L are modeled in this study: Jakarta Lower Control North (LN), Jakarta Lower Control Center (LC), Jakarta Lower Control East (LE), Jakarta Terminal West (TW), Jakarta Terminal East (TE), and Jakarta Terminal South (TS).Each sector has boundaries described by latitude-longitude, altitude/flight level, and radius from ATC head radar.Information about the boundaries can https://doi.org/10.31436/iiumej.v24i1.2223be accessed in Standard Operating Procedures Air Traffic Services Approach Control Service published by Airnav Indonesia Branch of Jakarta Air Traffic Service Center (JATSC) [21].The arrival traffic model has six trajectory profiles, and the departure traffic model has ten.Complete trajectory profiles and TCA sectors related to the Runway 25R and 25L air traffic model are shown in Fig .1.

Air Traffic Model
The air traffic model was built using MATLAB software by combining discrete-event and agent-based models.The wind speed model was added as an environmental element influencing the system.The wind speed consists of wind velocity and direction represents the weather condition.Each aircraft has a fixed parameter that will not change during simulation: aircraft type and trajectory profile based on SID/STAR.There are also dynamic parameters that will change during simulation; these parameters are: ▪ Position: In the form of local NED (North, East, Down) coordinates with a reference point at the NOKTA waypoint (X, Y, Z); ▪ Airspeed: Airspeed in the local NED direction (Vx, Vy, and Vz); ▪ Waypoint: present the waypoint to which the aircraft is headed; ▪ Heading: Heading aircraft relative to local north; ▪ Distance to waypoint: the distance of the aircraft to the next waypoint; ▪ Right of Way: priority of aircraft when heading/being on the same track; ▪ Conflict status: free from conflict or not; ▪ Resolution: selected conflict resolution mode (vectoring, speed control, or altitude control); ▪ TAS and GS: aircraft true airspeed and ground speed; ▪ Vertical speed: vertical aircraft speed when climbing (+) or descent (-); and ▪ RADAR radius: radius from RADAR.
Each trajectory has a unique profile based on STAR and SID published in Aeronautical Information Publication (AIP) [20].Each aircraft will move along the trajectory with the https://doi.org/10.31436/iiumej.v24i1.2223waypoint as profile guidance.The aircraft distance relative to the waypoint is obtained by the equation [22]: Xt, Yt, Zt: aircraft coordinates on each coordinate axis; Xwp, Ywp, Zwp: waypoint coordinates on each coordinate axis; dxt, dyt, dzt: distance to the waypoint on each coordinate axis; ds,t: aircraft distance relative to the waypoint.
After getting the distance relative to the waypoint using the above equation, then the heading angle can be calculated relative to the waypoint (  ) apply the equation: The aircraft heading angle relative to the waypoint is used to calculate the relative aircraft speed on each axis (Vx, Vy, Vz) using the equation below: Furthermore, it can be determined the position of the aircraft for each axis at a time (t + 1) through the equation: The separation between aircraft is maintained by using conflict detection and resolution models.It is necessary to calculate horizontal (dhor) and vertical (dver) separations between aircraft (a and b) to check whether the separation between aircraft is still safe (does not exceed the minimum limit) using the following equation: X, Y, and Z are the aircraft coordinates a and b on each coordinate axis.
After the separation between aircraft is known, whether the separation is still safe or if there has been a potential conflict (smaller than the specified minimum separation buffer) can be checked.Conflicts at TCA more often occur when the plane is heading to the merging point.If several aircraft experience conflict, it will be determined which aircraft gets the Right of Way (ROW) based on the closest distance to the merging point.The aircraft that gets the first ROW continues to fly following the specific trajectory without resolving conflict.The other aircraft should make specific maneuvers as part of conflict resolution.The flowchart of the aircraft's movement in the simulation model is shown in Fig. 2. The model has three conflict resolution modes: vectoring, airspeed control, and altitude control.

Flight Parameter Model
Flight parameter models were extracted from ADS-B data provided by FlightRadar24.The ADS-B data was collected from more than 38800 flights that departed and landed at Soekarno-Hatta International Airport, Jakarta [23].The models included five aircraft flight parameters: Airbus A320, Boeing B737, Airbus A330, Boeing B777, and Boeing B787.Machine learning was used to handle a large quantity of ADS-B data to identify the phase of the flight and the time when the aircraft flew across specific waypoints.One advantage of this technique is that ADS-B data can be efficiently and cost-effectively gathered over the internet.ADS-B data was combined with weather data from Aviation Meteorological Information System in Meteorological, Climatology, and Geophysical Agency (BMKG) to develop the flight parameter model [24].
The waypoints used for analysis were the waypoints flown by aircraft using Runway 25R and 25L for departing and arriving.The waypoint information was obtained from the Soekarno Hatta International Airport Terminal Chart published by the Indonesia DGCA [20].The K-Nearest Neighbor algorithms processed flight parameters from ADS-B data when the aircraft had the nearest position to specific waypoints within a radius of 1 NM.The altitude and the vertical speed were processed straight from the ADS-Data.The true airspeed was generated from the ground speed and wind speed data from weather data.
Three probability distribution functions (Normal Distribution, Beta Distribution, and Gamma Distribution) approached flight parameter models using maximum likelihood estimation.The best distribution was determined using Kolmogorov-Smirnov Test.The flight parameters for this research used the mean value from the models.Validation was carried out on previous research by comparing the estimated parameters with flight https://doi.org/10.31436/iiumej.v24i1.2223parameters from the Eurocontrol Aircraft Performance Database [25] and the reference parameter from the JAKARTA (WIII) Terminal Chart issued by the Indonesia Directorate General of Civil Aviation [26].Table 2 shows some of the aircraft's flight parameter models when flying by specific waypoints.

Air Traffic Complexity
The calculation of air traffic complexity at TCA begins by determining the indicators affecting the complexity and relevance of the airspace sector.After that, the weight of each indicator needs to be determined using the analytic hierarchy process (AHP) method.
The most mentioned indicators relevant to the TCA airspace sector can be selected based on Table 1.Hence, air traffic complexity measurement in this paper uses seven indicators: air traffic density (Nin), number of climbing (Nclb) and descending (Ndes) aircraft, aircraft type mixing (Ttyp), potential conflict control (crossing and overtaking conflict, Ttrf), aircraft speed difference (Tspd), and frequency of controller's coordination or communication (Tcom).The weight of each indicator was obtained using an analytic hierarchy process (AHP) method.

Waypoint
Altitude (feet) True Airspeed (knot) Vertical Speed (m/s) A320 B737 A330 B777 B787 A320 B737 A330 B777 B787 A320 B737 A330 B777 B787 The first step in the AHP method was to collect input data with pairwise comparisons of the indicators.Complexity indicators were paired with a rating that determined which indicator was more critical.This data was collected by a survey involving respondents from experts and practitioners.The questionnaire was created to assess the relative importance (weight value) of the target respondents, experts, and air traffic controllers from Airnav Indonesia, notably the Jakarta Air Traffic Services Center (JATSC).The questionnaire was filled out by respondents using online media.The second step was to average the input comparison values using the Row Geometric Mean Method (RGMM).The average ri value was determined using the following equation.
The comparison matrix A = aij with dimensions of N×N, N = 7 (number of indicators).The third step was calculating the first eigenvector of matrix A (Eigen 1 in the E1 matrix).The fourth step was calculating the second eigenvector (Eigen 2 in the E2 matrix).Then proceed to the fifth step, calculating the difference between E1 and E2.The sixth step was to assess the consistency of the respondent's answers in the following way.Table 3 shows the RI value for a given N value.After knowing the value of each indicator and its weight of importance, the general equation for the function of air traffic complexity at TCA can be calculated.
Air traffic complexity values were calculated for several scenarios.The scenarios were varied on the inter-arrival time (IAT) and the percentage of possible resolution types (speed control, altitude control, and vectoring).

RESULTS AND DISCUSSION
The simulation animation is used to validate the model by observing the model's behavior while the simulation is running [27].Some entities (aircraft) are observed moving from the time they enter the system to the time they leave the system to determine whether https://doi.org/10.31436/iiumej.v24i1.2223 the aircraft moves correctly according to the predetermined modeling concept and whether the conflict detection and resolution have been applied.
The animation observations show that the entity followed its arrival and departure trajectory according to the predetermined trajectory route.The data validation and operational graphics results also show that the model correctly implemented the applied conflict detection and resolution.No aircraft violated the minimum separation rules when the simulation ran in normal conditions.
The waypoint modeling for Soekarno-Hatta International Airport is based on local coordinates of NED with NOKTA as a reference point.The model is used to simulate the air traffic in the TCA, especially at JAKARTA TCA for Runway 25R and 25L.The model visualized 2-dimensional forms that moved for each unit of time.Flight parameters were estimated from ADS-B data for specific waypoints.
The movement of the aircraft in the simulation were repreented by a green dot when there was no conflict, a yellow dot when there was a potential conflict, and a red dot when there was a conflict with other aircraft.The simulation had twelve entry points and ran for 7200 units or the equivalent of 7200 seconds.There were three scenarios, each was running for just one resolution mode choice to solve the traffic conflict.Air traffic complexity was measured by calculating seven complexity indicators recorded during the simulation.Visualization of the simulation model is shown in Fig. 3.
A total of 119 questionnaire respondents from the JATSC controller provided information to determine which complexity indicator was more critical among the indicators that have been paired.The questionnaire results were then processed by AHP to obtain the weighting value for each indicator, as shown in Table 4.The level of consistency (Consistency Ratio) of the AHP was 0.72% which means that the answers were consistent (CR < 10%).This CR value indicated that the weighting values that were obtained could be used in the simulation.As shown in the table, the indicator that most influences complexity is the potential conflict control for crossing and overtaking conflict, with a weighting value of 38.87%.The indicator with the smallest weighting value https://doi.org/10.31436/iiumej.v24i1.2223 is the number of climbing aircraft (8.99%).Thus, the general function for air traffic complexity at JAKARTA TCA can be written as follows.Function (18) shows that the number of descending aircraft is more critical than the number of climbing aircraft in influencing air traffic complexity.It is also shown that the aircraft type mix is more important than the aircraft speed difference.This result is different from Diaconu et al. (2014) in which the climbing was more important than the descent, and the aircraft speed difference was more important than the aircraft type mix [4].The controller's significant preference with the same indicator may vary for different air traffic service units.So it is necessary to analyze the weighting value of the complexity indicator for related air traffic service units before measuring the air traffic complexity in specific airspace sectors.
Controller task load is represented by controller communication time in the function through the frequency of the controller's coordination or communication (  ).The complexity rate will increase the more frequently the controller coordinates or communicates with the pilot and another controller.The weighting of communication time in the complexity function is more critical than the number of climbing aircraft (  ) and Aircraft Types Mixing (  ).In addition to task load, other factors such as equipment capability, individual preferences, and cognitive controller strategies are required to obtain a complete picture of the correlation between complexity and workload [28].
The simulation run for ten repetitions with aircraft type mix is a 9:1 ratio for Medium type (Boeing B737 and Airbus A330) and Heavy type (Airbus A330, Boeing B777, and Boeing B787).When entering the arrival point, time separation is set to 4 minutes, and departure is about 6 minutes.This gives high traffic density to the model.The minimum separation is 5 NM with a buffer of 15 NM to solve the potential conflict.A graphic of complexity measurement from a simulation (altitude control mode only) for each TCA sector is shown in Fig. 4.
From Fig. 4, Jakarta Lower Control North and Jakarta Terminal West have a high rate of complexity relative to the other sectors.Jakarta Terminal East has the lowest complexity rate compared with the others on Runway 25R and 25L operation.The model can show the complexity rate comparison between sectors.It can be used to assess what sector has a higher rate of complexity for a particular runway operation, and the management should take some action to balance it.Simulation results of air traffic complexity measurement and the number of potential conflicts for specific resolution mode scenarios in each sector are shown in Fig. 5 and 6.  Figure 5 shows that the speed control mode assigns a higher complexity rate for all sectors except Jakarta Lower Control North.For lower rate complexity, vectoring mode gives the lowest complexity than the other resolution mode, as seen in Jakarta Lower Control North, Jakarta Lower Control Centre, and Jakarta Terminal South.Altitude control mode gives a higher rate of complexity and more potential conflict if applied to the sector with many departure trajectories like Jakarta Lower Control North.Fig. 5 and 6 elaborate that speed control to solve the conflict increases complexity and creates more potential conflict for almost all sectors.Vectoring mode gives the least potential conflict than the other modes except on Jakarta terminal West.
The simulation result can compare the complexity between sectors in a TCA and its effect on potential conflicts.However, the model cannot yet determine how this complexity affects aviation safety risks.As we know, complexity will affect the controller's workload level [5].Representation of the human factor is needed to measure the controller workload, and its effect on safety risk factors can be observed.Future research needs methods to represent the human factor in a simulation model, including adding a controller to the simulation (human in the loop simulation) [16] or developing a controller workload model [29].

CONCLUSION
The weighting value of seven air traffic complexity indicators has been calculated using the AHP method in this study.The indicator that most significantly affects the air traffic complexity is the potential conflict control.The consistency of respondents' answers is less than 10%, indicating that these results are consistent.It can determine the air traffic complexity rate in the JAKARTA TCA model.
The simulation has been run to measure the air traffic complexity of JAKARTA TCA in a high-density situation.The simulation result explains that the resolution mode selection

Fig. 2 :
Fig. 2: Flow chart of aircraft movements in the simulation model.
a. Calculate the Weighted Sum Vector (WSV) by multiplying the rows of matrix A by matrix E1.b. Divide each element of the WSV matrix by each element of the E1 matrix to obtain the Consistency Vector (CV).c.Calculate the lambda (λ) by averaging CV and calculating the Consistency Index (CI) using the following equation.CI by the Random Consistency Index to get the Consistency Ratio (CR) (RI).

Fig. 3 :
Fig. 3: Visualization of the simulation model with aircraft movement representation.

Fig. 5 :
Fig. 5: Values of air traffic complexity on sectors for specific resolution mode's scenario.

Fig. 6 :
Fig. 6: Number of potential conflicts on sectors for specific resolution mode's scenario.
(g) Jakarta Lower Control North (h) Jakarta Lower Control Center (h) Jakarta Lower Control East (i) Jakarta Terminal West (j) Jakarta Terminal East (k) Jakarta Terminal South https://doi.org/10.31436/iiumej.v24i1.2223influences the complexity rate and potential conflicts.The simulation model requires further development by representing the human factor in the model so the model can be used to analyze safety risks.

Table 1 :
Air traffic complexity indicators summary

Table 2 :
Flight parameter models for each type of aircraft when flying by waypoints

Table 4 :
Weighting value of each complexity indicator